Thbs4 labels type-B cells at the neurogenic niche in the SVZ.

(a) Brain schematic drawing showing Thbs4 labelling throughout the mouse brain. Thbs4 was mainly found in the SVZ and RMS in the adult mouse telencephalon. Arrows point to unspecific Thbs4 staining in cerebellum and ventral tegmental area. (b) Thbs4 positive cells in the SVZ showed different morphologies. (c) Proportion of Thbs4-positive cells among the total cell population at the SVZ. (d) Classification of Thbs4-positive cells in the SVZ according to their NSC morphology. (e) Electroporation of pThbs4-GFP at P1 revealed Thbs4 expression in the post-natal dorsal SVZ. (f) Quantification of Thbs4-reporter expression showing the proportion of NSC expressing Thbs4 postnatally in the dorsal SVZ. (g) Thbs4-GFP reporter expression was higher expressed in the rostral SVZ. BV: Blood vessel; LV: Lateral ventricle. Scale bar = 10 µm (b) and 50 µm (e). Bars represent mean ± SEM. *p<0.05 and **p<0.01, Tukey post hoc test (after one-way ANOVA was significant at p < 0.05).

Thbs4 increase in the SVZ after brain ischemia.

(a) Representative images of Thbs4 (green), Nestin (red) and DCX (magenta) markers in physiological (left) and 15 days post-injury (dpi) SVZ (right). (b,c) Active proliferating NSCs and neuroblast decrease at 7 and 15 dpi in the SVZ, as shown by Nestin (b) and DCX (c) levels respectively. decreased 7 and 15 dpi in the SVZ. (c) DCX marker (neuroblasts) decreased 7 and 15 dpi in the SVZ. (d) Thbs4 expression increased in the SVZ after the ischemic insult over time. (e) Representative images of Nestin and Thbs4 protein in sham and 15 dpi SVZ by western blot. (f,g) Quantification of mean gray value (MGV) levels from (e), normalized to GAPDH, show Thbs4 (f) increase and Nestin decrease (g) in the 15 dpi SVZ. Scale bar = 100 µm in (a). n = 6 (b, c, d) and 3 (e, f, g) for each condition. Bars represent mean ± SEM. *p < 0.05 and **p < 0.01; two-tail Student’s t test (sham vs. 15ddMCAO) and Tukey post-hoc test (after one-way ANOVA was significant at p < 0.05). See also Supplementary Figures 1 and 2.

Thbs4-positive cells derive from B-type NSC in the SVZ after brain ischemia.

(a) Experimental design: chronic BrdU (1% in water) treatment was administered for two weeks. BrdU only persisted in B-type NSCs 30 days after treatment. Animals were injected three times with IdU (50 mg/kg) to label proliferating cells the day before euthanasia. (b,c) BrdU (b) and IdU (c) positive cells at the SVZ. Only BrdU cells are significantly increased at 30 dpi. (d) Representative confocal images of Thbs4, GFAP and BrdU levels in sham (top) and 30 dpi mice (bottom). (e,f) Thbs4/GFAP/BrdU positive cells (slow proliferative cells) increased in the SVZ 30 dpi (e), while no changes were found for Thbs4/GFAP/IdU positive cells (fast proliferative cells). Scale bar = 10 µm (d). n = 4 (sham) and 3 (MCAO). Bars represent mean ± SEM. *p < 0.05 by two-tail Student’s t-test (sham vs. 30dd MCAO). See also Supplementary Figure 3.

Ischemia-induced Thbs4 astrocytes migrate from the SVZ to ischemic areas.

(a) Experimental design: Ai14, ROSA26-CAG-tdTomato transgenic mice were electroporated in the dorsal SVZ at P1. pCAGGSx-CRE plasmid induced expression of tdTomato (tdTOM) in dorsal NSCs. MCAO was performed after 3-4 months in the same electroporated hemisphere. Mice were analyzed 7, 30 and 60 dpi. (b) Quantification of tdTOM-positive cells shows a decrease in the dorsal SVZ after brain ischemia. (c) Representative images of tdTOM positive cells in sham (top) and 60 dpi SVZ (bottom). (d) Quantification of tdTOM positive cells in the whole brain show incremental increase outside of the SVZ after brain ischemia. (e) Representative images of tdTOM expression in sham (left) and 60 dpi mice (right). (f) Representative image of Thbs4/tdTOM positive cells in the dorsal SVZ. (g) Quantification shows an increase of Thbs4/tdTOM-positive cells in the SVZ 30 and 60 dpi. (h) Proportion of Thbs4-positive cells in the tdTOM pool increases in the SVZ 30 and 60 dpi. (i) Around 50% of tdTOM-positive cells expressed Thbs4 in the infarcted tissue 60 dpi. (j) Representative image of Thbs4 and tdTOM expression near damaged area. Scale bar = 50 µm (c), 10 µm (f) and 100 µm (g). n = 6 animals per condition. Bars represent mean ± SEM. *p < 0.05, **p < 0.01 and ***p < 0.001 by Chi-square test (h) and Tukey posthoc test (after one-way ANOVA was significant at p < 0.05). See also Supplementary Figure 4.

Ischemia-induced Thbs4 astrocytes accumulate at the glial scar.

(a) Glial scar could be observed at 30 dpi by Nestin immunofluorescence. (b) Classical TTC technique underestimated ischemic regions compared with Nestin immunofluorescence. (c) Representative images of TTC staining I infarcted brains. (d) Representative confocal images showing the time course of Nestin expression at 3, 7, 15 and 30 dpi. (e) Nestin glial scar was established by 15 dpi. Thbs4 astrocytes arrived at the glial scar later, by 30 dpi. (f) Fluorescence profile of Thbs4, GFAP and Nestin marker from core to penumbra areas shown Thbs4 was mainly located in the borders of the Nestin-positive glial scar. (g) Representative images of Thbs4 (green), GFAP (white) and Nestin (red) markers in sham (top) and 30 dpi mice (bottom). Scale bar = 50 µm (g). n = 3 animals per condition. Bars/lines represent mean ± SEM. *p < 0.05 and **p < 0.01 by two-tail Student’s t test (TTC vs. Nestin) and by Dunnett posthoc test (after two-way ANOVA was significant at p < 0.05). See also Supplementary Figure 5.

Thbs4-positive astrocytes produce hyaluronan at the glial scar after MCAO.

(a) Representative images of hyaluronic acid (HA) labelled with HABP in the contralateral (1) and ipsilateral hemisphere (2). Only perineuronal nets are clearly visible in the healthy tissue, while a dense interstitial matrix is observed in ischemic areas (b) Cumulative distribution of skeletonized HA signal in sham and infarcted regions of 7, 15 and 30 dpi mice (p<0.0001). (c) Quantification of fractal dimension showed increased extracellular matrix interconnectivity in the infarcted cortex and striatum (Ctx&Str) 30 dpi. (d) Representative confocal images of Thbs4 (green) and HABP (red) in the cortex and striatum of sham and 30 dpi mice. Notice the HA labelling on the surface of Thbs4-positive cells. (e) Fluorescence intensity profile of Thbs4, GFAP and HABP markers from lesion core to penumbra areas. Thbs4 expression was mainly restricted to the borders of the HA glial scar 30 dpi. ((f) Representative confocal images of Thbs4 and HABP markers in sham and 30 dpi mice (top). Examples of area occupied by HA (red) in the membrane mask of Thbs4 positive astrocytes in sham and 30 dpi mice (bottom), showing increased HA synthesis after MCAO. (g) Quantification of membrane-bound HA showed increased HA along Thbs4-positive membrane after brain ischemia (corpus callosum: pink; infarcted cortex: blue; infarcted striatum: green). Scale bar = 10 µm (e). n = 6 animals per condition. *p < 0.05 and ****p < 0.0001 by Krustal Wallis test and two-way ANOVA (significant at p < 0.05). See also Supplementary Figures 6 and 7.

Hyaluronan accumulation is sufficient but not exclusive to recruit Thbs4 astrocytes.

(a) Representative images of HA accumulation caused by viral-mediated overexpression of HAS2 from naked mole rat. (b) Experimental design: HAS2 viral injections were performed 2 weeks before MCAO. Animals were sacrificed 30 days after MCAO. We used standard GFP virus as control. Two groups of HAS2 injected mice were performed. Only one of them was subjected to MCAO. (c) FLAG-positive cells (HAS2 virus tag) did not show changes in either HAS2 control (HAS2Control) or HAS2-MCAO group (MCAO contralateral hemisphere). (d,e,f) Representative images of HABP, GFP and Thbs4 markers in GFP control virus (d), HAS2 control virus (e) and HAS2 MCAO group (f). (g) Thbs4 expression increased after HAS2 viral-mediated overexpression. However, Thbs4 was upregulated in the ischemic hemisphere when MCAO was performed together with the HAS2 viral infection (MCAO ipsi). GFP control virus did not induce increase of Thbs4. (h) HABP was used as a control for HA accumulation. Thbs4 astrocytes in HAS2 MCAO group increased in the ischemic hemisphere even though HA accumulation was the same in both hemispheres (HAS2 and MCAO hemisphere), suggesting other ischemia-induced signals recruit Thbs4 astrocytes to infarcted areas. Scale bar = 100 µm (a). n = 6 animals per condition. *p < 0.05, **p < 0.01 and ***p < 0.001 by two-tail Student’s t test (HAS2Control vs. MCAO Contra) and Tukey posthoc test (after one-way ANOVA was significant at p < 0.05).

HA degradation in the SVZ correlates with Thbs4 response to brain ischemia.

(a) Representative image of Thbs4 and HA (HABP) markers in the SVZ 30 dpi. (b) Representative images of Thbs4 and HA in the sham and 15, 30 dpi SVZ. (c, d) HA signal did not change in the whole SVZ after MCAO (c) but decreased drastically in the dorsal SVZ (d). Thbs4 expression increased overall in the SVZ after MCAO (c), with a marked increase at 30 dpi in the dorsal SVZ (d), just after the HABP decrease. (e) Representative HABP confocal (top) and skeletonized images (bottom) in the dorsal SVZ of sham, 15 and 30 dpi mice. (f) Cumulative distribution of HA skeletons showed a decrease in HA-cables length only in the dorsal SVZ at 7 and 15 dpi. (g) Interconnectivity of the extracellular matrix decreased by 7 dpi in the dorsal SVZ, as measured by fractal dimension. (h) Experimental design for qPCR experiment: 3/4-month mice were subjected to MCAO. Fresh SVZ tissue was extracted 15 and 30 days after MCAO and processed for RT-qPCR analysis. (i) Hyaluronan degradation genes (Hyal1, 2 and 3) increased by 15 and 30 dpi in the SVZ, whereas hyaluronan synthase genes (HAS1 and HAS2) were overexpressed later 30 dpi in the SVZ. Thbs4 and Hif1a genes were also increased by 15 and 30 dpi in the SVZ. Scale bar = 100 µm (a). n = 6 (c, d, e and g) and 3 (i). *p < 0.05, **p < 0.01 and ***p < 0.001 by Krustal Wallis test and Tukey posthoc test (after one-way ANOVA was significant at p < 0.05).

Primary antibodies used for in vivo immunofluorescence protocol

Primer sequences for qPCR

Characterization of the Middle Cerebral Artery Occlusion (MCAO) mouse model of brain ischemia.

Related to Figure 2. (a) Scheme of the procedure: 10 mm silicon-coated filament was inserted through an external artery and re-directed to the internal one. Ischemia was induced after 60 minutes of occlusion. After that, the filament was removed, and animals were observed every day before euthanasia. (b) Volume of infarction was observed by cresyl violet staining. (c) Neuronal death was observed by NeuN immunofluorescence. (d) Around 50% of ischemic mice survived at 28 dpi. (e) Ischemic animals ameliorated neurological symptoms overtime. (f) Weight was measured as controls for mice healthcare. n = 17 (d, e, f). **p < 0.01 and ****p < <0.0001 by survival curve test and Tukey posthoc test (after one-way ANOVA was significant at p < 0.05).

Cell proliferation in the SVZ after brain ischemia.

Related to Figure 2. (a) Experimental design: three BrdU (50mg/kg) injections were performed the day before MCAO protocol. Animals were sacrificed 24 hours after the brain stroke model. (b) BrdU positive cells increased in the SVZ 24 hours after MCAO. (c) Representative image of BrdU positive cells in the ipsilateral (left) and contralateral (right) hemisphere 24 hours after MCAO. (d) Ki67 marker was also used to analyze cell proliferation in wholemount preparations of the SVZ. (e) Ki67 positive cells increased in the SVZ 24 hours after MCAO. (f) The Cleaved-Caspase 3 marker increased in the SVZ 24 hours after MCAO. (g) Cleaved-caspase 3 marker was used to assess apoptosis in the SVZ after brain ischemia. Scale bar = 15 µm (g). n = 5 (sham) and 4 (MCAO). Bars represent mean ± SEM. *p < 0.05, **p < 0.01 and ***p < 0.001 by two-tail Student’s t test (sham vs. 24h MCAO).

Thbs4 astrocytes derive from slow proliferative cells after brain ischemia.

Related to Figure 3. (a) Slow proliferative cells were quantified in the lateral (grey), ventral (orange), medial (yellow) and dorsal (blue) SVZ. We only observed an increase of Thbs4/GFAP/BrdU positive cells in the dorsal SVZ 30 dpi. (b) Thbs4/GFAP/IdU positive cells did not show changes in the lateral (grey), ventral (orange), medial (yellow) and dorsal (blue) SVZ after brain ischemia. n = 4 (sham) and 3 (MCAO). **p < 0.01 by by two-tail Student’s t test (sham vs. 30dd MCAO).

Ischemic-induced Thbs4 astrocytes migrate preferentially to caudal infarcted areas.

Related to Figure 4. (a) Representative image of tdTOM positive cells in the OB three weeks after electroporation (3wpi). (b) tdTOM-positive cells decreased in the rostral, intermediate and caudal SVZ 7, 30 and 60 dpi. (c) Thbs4/tdTOM positive cells increased in rostral and intermediate SVZ 30 and 60 dpi. (d) Expression of tdTomato increased in intermediate and caudal brain areas 30 dpi. (e) Representative images of tdTOM expression in rostral (left), intermediate (middle) and caudal (right) brain areas of sham (top) and 60 dpi mice (bottom). (f) tdTOM positive cells increased in the infarcted cortex (Ctx) and striatum (Str) whereas decreased in the dorsal SVZ (SVZ). No changes were observed in the corpus callosum (CC). (g) tdTOM positive cells only increased in the infarcted cortex and striatum at intermediate and caudal brain regions. Scale bar = 100 µm (a) and 1000 µm (e). n = 6 animals per condition. Bars represent mean ± SEM. *p < 0.05, **p < 0.01 and ***p < 0.001 by Dunnett/Tukey posthoc test (after two-way ANOVA was significant at p < 0.05).

SVZ NSC stop producing newborn neurons to send newborn astrocytes to the ischemic areas.

Related to Figure 5. (a) Thbs4 expression in the corpus callosum (cc), infarcted cortex (Ctx) and infarcted striatum (Str) of sham and 7 and 30 dpi mice. (b) Thbs4 increased by 30 dpi in the corpus callosum (cc) and 60 dpi in the infarcted cortex (Ctx) and infarcted striatum (Str). (c) Expression of Thbs4 increased in total GFAP population after 15, 30 and 60 dpi. (d) Experimental design: chronic BrdU (1% in water) treatment was administered for two weeks. BrdU only persisted in slow proliferative NSC (Codega et al., 2014). DCX and BrdU positive cells were measured in the OB layers 7, 15 and 30 dpi. (e) Representative images of BrdU and DCX marker in sham (left), 15 (middle) and 30 (right) dpi mice OB. (f) DCX positive cells decreased 15 and 30 dpi in the RMS, granular OB layer (GrL) and glomerular OB layer (GL). (g) BrdU positive cells only decreased 30 dpi in the glomerular OB layer (GL). Double DCX/BrdU positive cells significative decreased in the RMS 30 dpi. In (f) and (g), values are normalized to total DAPI nuclei (100%). Scale bar = 500 µm (a). n = 6 animals per condition. Lines represent mean ± SEM. *p < 0.05, **p < 0.01 and ***p < 0.001 by Chi-square test and Tukey posthoc test (after two-way ANOVA was significant at p < 0.05).

Thbs4 astrocytes internalize HA in the infarcted areas.

Related to Figure 6. (a) Representative images of Thbs4/GFAP (“newborn astrocytes”, top) and GFAP-only (“local astrocytes”, bottom) cells in the infarcted areas 30 dpi. (b) Thbs4-positive astrocytes internalized more HA compare to local astrocytes only labelled with GFAP marker. (c) However, the number of HA spots internalized was higher in local astrocytes compared with Thbs4 astrocytes, suggesting that Thbs4 internalize HA in fewer vesicles. (d) Representative images of HABP (red), Thbs4 (green) and CD44 (white) in the ischemic areas. CD44 is well-known for its role in HA degradation and ECM remodeling. (e) Thbs4 positive astrocytes expressed a high rate of CD44 receptors so much in corpus callosum (CC), infarcted cortex (Ctx) as in infarcted striatum (Str). Scale bar = 10 µm (c, f). n = 5 (b, d and e) and 3 (g) animals per condition. Bars represent mean ± SEM. *p < 0.05 and **p < 0.01 by Tukey posthoc test (after two-way ANOVA was significant at p < 0.05) and two-tail Student’s t test (Thbs4 vs GFAP).

In vitro NSC-derived Thbs4 cells synthetize HA after a hypoxic-ischemic condition.

Related to Figure 6. (a) Experimental design: P4 pup rats were sacrificed, and SVZ NSC were extracted and cultured in the proliferative medium for one week. After that, neurospheres were disaggregated and seeded isolated in a culture dish. The following day, oxygen and glucose deprivation protocol (OGD) was performed. Cells were fixed 7 days after culture them. (b) Representative images of in vitro Thbs4 positive cells (green) and HA (red). (c) HA spots were measured inside and outside Thbs4 positive cells. (d) HA spots increased inside Thbs4 positive astrocytes after OGD whereas (e) decreased in the extracellular space. However, when NSC were exposed to hyaluronidases in order to remove extracellular HA, Thbs4 positive cells decreased internalization of HA spots (f) and increased extracellular HA spots (g). (h) Neurosphere-derived NSC (NSPH1) internalized more HA spots compare with Iba1-positive cells. (i) Metalloproteinase activity was higher in OGD NSC cultures compared to control condition. Scale bar = 10 µm (b). n = 15 samples per condition. Bars represent mean ± SEM. *p < 0.05, **p < 0.01 and ***p < 0.001 by two-tail Student’s t test (CTRL vs. OGD; NSPH1 vs. Iba1+).